Reproducibility and its confounders of CMR feature tracking myocardial strain analysis in patients with suspected myocarditis

Abstract

Objectives: Cardiovascular magnetic resonance feature tracking (CMR-FT) is an emerging technique for assessing myocardial strain with valuable diagnostic and prognostic potential. However, the reproducibility of biventricular CMR-FT analysis in a large cardiovascular population has not been assessed. Also, evidence of confounders impacting reader reproducibility for CMR-FT in patients is unknown and currently limits the clinical implementation of this technique.

Methods: From a dual-center database of patients referred to CMR for suspected myocarditis, 125 patients were randomly selected to undergo biventricular CMR-FT analysis for 2-dimensional systolic and diastolic measures, with additional 3-dimensional analysis for the left ventricle. All image analysis was replicated by a single reader and by a second reader for intra- and inter-reader analysis (Circle Cardiovascular Imaging). Reliability was tested with intraclass correlation (ICC) tests, and the impact of imaging confounders on agreement was assessed through multivariable analysis.

Results: Left and right ventricular ejection fractions were reduced in 34% and 37% of the patients, respectively. Good to excellent reliability was shown for 2D (all ICC > 0.85) and 3D (all ICC > 0.70) peak strain and early diastolic strain rate for both ventricles in longitudinal orientation as well as circumferential orientations for the left ventricle. An increased slice number improved agreement while the presence of pericardial effusion compromised diastolic strain rate agreement, and arrhythmia compromised right ventricular agreement.

Conclusion: In a large clinical cohort, we could show CMR-FT yields excellent inter-reader and intra-reader reproducibility. Multi-parametric CMR-FT of the right and left ventricles appears to be a robust tool in cardiovascular patients referred to CMR.

Clinicaltrials: gov Identifier: NCT03470571, NCT04774549. Key Points • Cardiovascular magnetic resonance feature tracking (CMR-FT) is an emerging technique to measure myocardial strain in cardiovascular patients referred for CMR; however, the evaluation of its reproducibility in a large cohort has not yet been performed. • In a large clinical cohort, CMR-FT yields excellent inter-reader and intra-reader reproducibility for both left and right ventricular systolic and diastolic parameters. • Arrhythmia and pericardial effusion compromise agreement of select FT parameters, but poor ejection fraction does not.

Keywords: Magnetic resonance imaging; Myocarditis; Reproducibility of results; Ventricular dysfunction.

Fig. 1

Feature tracking analysis. Feature tracking is performed on short-axis (a) and long-axis cinés (b) for 2D analysis, which can be constructed to form a 3D model (c). d A typical strain (blue) and strain rate (green) curve are shown for the longitudinal orientation, marking the key measurements; PS: peak strain, TTP: time to peak strain, sSR: systolic strain rate, edSR: early diastolic strain rate, and adSR: late diastolic strain rate

Fig. 2

Intraclass correlation coefficients. Inter-reader (square) and intra-reader (diamond) intraclass correlation coefficients (ICC) and 95% confidence intervals demonstrated significant reliability (all p < 0.05). Green zone: excellent reliability (≥ 0.90), blue zone: good reliability (0.75–0.90). LV: left ventricle, RV: right ventricle

Fig. 3

Examples of agreement. a Strain and strain rate curves are similar between all three reads and demonstrating excellent agreement in this 30-year-old patient with a resting heart rate of 65 bpm and a late gadolinium extent of 27%. b The second case shows poorer agreement, especially in the right ventricle (RV) in comparison to the left (LV) from a 41-year-old patient with a resting heart rate of 109 bpm who had 5% late gadolinium enhancement